// learn.shawon.ch / evidence-desk / operating-filters PERSONAL
← Evidence desk

Evidence desk · Personal

Operating filters

What this is

The priors I bring to every verdict, set before scoring any single claim. Two layers: durable filters — who I am biologically, which raises or lowers the bar on whole categories of claim — and phase logic — the fact that the same supplement can be skip-now, great-later. Every entry on this desk is read through both.

Durable filters

These set the priors before a claim is even scored. A claim that lands in a high-priority lane (cardiometabolic, given the profile below) clears a lower bar to act once it's validated, because high baseline risk means a bigger absolute benefit. A claim that's redundant with what a drug already does, or that collides with a medication, gets deprioritized or routed to a clinician.

The lenses every claim passes through
LensWhat it changes
Metabolic — South Asian, insulin-resistant (HOMA-IR 3.65), Pattern B / small-LDL, ApoB target <70Cardiometabolic, lipid, and glucose claims get higher priority and a lower bar to act once validated (high baseline risk → larger absolute benefit). Check redundancy with tirzepatide first.
On tirzepatide (Zepbound)Glucose-lowering claims are often redundant. GI-loading supplements get deprioritized. Appetite suppression makes protein hard → protein-sparing matters.
Medications — lisinopril, Adderall XR, finasteride, oral minoxidilScreen vasoactive / BP claims for additive hypotension; screen CYP3A4 / P-gp and vitamin-C / pH timing against Adderall. Anything touching a med routes through the prescriber.
Genetics — VDR reduced-function, GSTP1 ~60%, TCF7L2, CYP2D6 pending WGSD3 need is real; glutathione-support has a rationale; glycemic control is a permanent priority; metabolizer-dependent claims stay uncertain until the WGS lands.
Skin — Fitzpatrick ~4.5, pigmentation work, tretinoin on boardJudge skin / antioxidant claims against pigmentation + photoprotection. Collagen is skin/tendon support, not protein.
Sleep — low-to-moderate OSA, low REM / durationSeparate "sleep-quality supplement" from "treats apnea." Apnea is mechanical — no capsule fixes it; the real levers are weight loss + the OSA pathway (PSG, CPAP/MAD).
System fit — ADHD, pill fatigue, system-not-moodComplexity is a real cost → bias to a minimal, high-leverage stack. Prefer claims with measurable, trackable (DEXA / lab) outcomes over "feel."

Phase logic

The same claim can be skip-now and great-later. Which phase I'm in changes what's worth taking and — just as important — whether I can even attribute an effect.

Phase 1 · Cut & Recomp (now → ~157 lb)

Lens: aggressive fat loss on a GLP-1; muscle retention is the prize; attribution is muddied (drug + deficit running at once).

Favor: complete protein (whey), creatine, resistance training, psyllium (LDL + satiety + regularity), magnesium repletion, D3, omega-3 maintenance, skin/connective-tissue support.

Skip / defer: anything redundant with tirzepatide (glucose / "metabolism" boosters), GI-competing pills, incomplete protein counted as protein, placebo-loaded "feel" supplements (a deficit inflates placebo), most new n-of-1s — you can't attribute cleanly now.

Attribution rule: don't start multiple things at once; judge on DEXA / labs / logged lifts, never on feel.

Phase 2 · Lean Build (~157 → 170 lb, optional)

Lens: surplus, hypertrophy; if the GLP-1 is off, appetite and attribution both clear → the best window for clean n-of-1s.

Favor: creatine (continues), easy complete protein, performance ergogenics now justifiable (citrulline, beta-alanine, caffeine timing), recovery / sleep support for the higher training stress.

Revisit: items skipped in the cut for redundancy or GI load; longevity items stay watch-only unless human hard-outcome data has matured.

Phase 3 · Sustain / Healthspan (held at goal)

Lens: maintenance + long-term health; body-comp goal met → focus shifts to cardiometabolic durability, cognition, joints, longevity.

Favor: validated-surrogate maintenance (ApoB / lipids, glucose, BP), bone/joint preservation, cognitive maintenance, antioxidant / glutathione support (GSTP1), and longevity compounds that have since climbed the evidence ladder.

The phase point: a compound with no body-comp ROI and noisy attribution in the cut (skip) can be a legitimate sustain-phase n-of-1 once the goal is healthspan, the stack is stable, and attribution is clean. Watch-only graduates to provisional-try here, never earlier.

The decision rule

Once a claim clears the filters and the phase lens, the act/skip call is the same four-box rule from is it strong enough to act?:

Evidence × cost/risk
Low cost / riskHigh cost / risk
Strong evidenceDo it, permanently.Do it if the upside justifies the downside — clinician-gated.
Weak evidenceReasonable to try as a pre-set n-of-1.Don't. This is where hype lives.

Asymmetry is half the decision: cheap, safe, reversible + might-help → try even on weak evidence; expensive, risky, or irreversible + might-help → wait even when tempted. An honest n-of-1 changes one variable, fixes an objective metric + timeframe + stop rule in advance, expects regression to the mean, and never judges on feel — best run in Build or Sustain, not the Cut.

Quarterly review

Every quarter, walk the stack: has anything drifted above its evidence? Is anything watch-only now backed by human hard-outcome data (promote it)? Has anything in the permanent column been demoted by newer evidence? Did I add something multi-variable and break attribution? Did I act on a relative or surrogate claim without checking the absolute number or validation?

Learn · Shawon Chowdhury · personal operating filters · not medical advice